Re-implement md32_common.h [make it simpler!] and eliminate code rendered
[openssl.git] / crypto / sha / sha256.c
1 /* crypto/sha/sha256.c */
2 /* ====================================================================
3  * Copyright (c) 2004 The OpenSSL Project.  All rights reserved
4  * according to the OpenSSL license [found in ../../LICENSE].
5  * ====================================================================
6  */
7 #include <openssl/opensslconf.h>
8 #if !defined(OPENSSL_NO_SHA) && !defined(OPENSSL_NO_SHA256)
9
10 #include <stdlib.h>
11 #include <string.h>
12
13 #include <openssl/crypto.h>
14 #include <openssl/sha.h>
15 #include <openssl/opensslv.h>
16
17 const char *SHA256_version="SHA-256" OPENSSL_VERSION_PTEXT;
18
19 int SHA224_Init (SHA256_CTX *c)
20         {
21         memset (c,0,sizeof(*c));
22         c->h[0]=0xc1059ed8UL;   c->h[1]=0x367cd507UL;
23         c->h[2]=0x3070dd17UL;   c->h[3]=0xf70e5939UL;
24         c->h[4]=0xffc00b31UL;   c->h[5]=0x68581511UL;
25         c->h[6]=0x64f98fa7UL;   c->h[7]=0xbefa4fa4UL;
26         c->md_len=SHA224_DIGEST_LENGTH;
27         return 1;
28         }
29
30 int SHA256_Init (SHA256_CTX *c)
31         {
32         memset (c,0,sizeof(*c));
33         c->h[0]=0x6a09e667UL;   c->h[1]=0xbb67ae85UL;
34         c->h[2]=0x3c6ef372UL;   c->h[3]=0xa54ff53aUL;
35         c->h[4]=0x510e527fUL;   c->h[5]=0x9b05688cUL;
36         c->h[6]=0x1f83d9abUL;   c->h[7]=0x5be0cd19UL;
37         c->md_len=SHA256_DIGEST_LENGTH;
38         return 1;
39         }
40
41 unsigned char *SHA224(const unsigned char *d, size_t n, unsigned char *md)
42         {
43         SHA256_CTX c;
44         static unsigned char m[SHA224_DIGEST_LENGTH];
45
46         if (md == NULL) md=m;
47         SHA224_Init(&c);
48         SHA256_Update(&c,d,n);
49         SHA256_Final(md,&c);
50         OPENSSL_cleanse(&c,sizeof(c));
51         return(md);
52         }
53
54 unsigned char *SHA256(const unsigned char *d, size_t n, unsigned char *md)
55         {
56         SHA256_CTX c;
57         static unsigned char m[SHA256_DIGEST_LENGTH];
58
59         if (md == NULL) md=m;
60         SHA256_Init(&c);
61         SHA256_Update(&c,d,n);
62         SHA256_Final(md,&c);
63         OPENSSL_cleanse(&c,sizeof(c));
64         return(md);
65         }
66
67 int SHA224_Update(SHA256_CTX *c, const void *data, size_t len)
68 {   return SHA256_Update (c,data,len);   }
69 int SHA224_Final (unsigned char *md, SHA256_CTX *c)
70 {   return SHA256_Final (md,c);   }
71
72 #define DATA_ORDER_IS_BIG_ENDIAN
73
74 #define HASH_LONG               SHA_LONG
75 #define HASH_CTX                SHA256_CTX
76 #define HASH_CBLOCK             SHA_CBLOCK
77 /*
78  * Note that FIPS180-2 discusses "Truncation of the Hash Function Output."
79  * default: case below covers for it. It's not clear however if it's
80  * permitted to truncate to amount of bytes not divisible by 4. I bet not,
81  * but if it is, then default: case shall be extended. For reference.
82  * Idea behind separate cases for pre-defined lenghts is to let the
83  * compiler decide if it's appropriate to unroll small loops.
84  */
85 #define HASH_MAKE_STRING(c,s)   do {    \
86         unsigned long ll;               \
87         unsigned int  n;                \
88         switch ((c)->md_len)            \
89         {   case SHA224_DIGEST_LENGTH:  \
90                 for (n=0;n<SHA224_DIGEST_LENGTH/4;n++)  \
91                 {   ll=(c)->h[n]; HOST_l2c(ll,(s));   } \
92                 break;                  \
93             case SHA256_DIGEST_LENGTH:  \
94                 for (n=0;n<SHA256_DIGEST_LENGTH/4;n++)  \
95                 {   ll=(c)->h[n]; HOST_l2c(ll,(s));   } \
96                 break;                  \
97             default:                    \
98                 if ((c)->md_len > SHA256_DIGEST_LENGTH) \
99                     return 0;                           \
100                 for (n=0;n<(c)->md_len/4;n++)           \
101                 {   ll=(c)->h[n]; HOST_l2c(ll,(s));   } \
102                 break;                  \
103         }                               \
104         } while (0)
105
106 #define HASH_UPDATE             SHA256_Update
107 #define HASH_TRANSFORM          SHA256_Transform
108 #define HASH_FINAL              SHA256_Final
109 #define HASH_BLOCK_DATA_ORDER   sha256_block_data_order
110 void sha256_block_data_order (SHA256_CTX *ctx, const void *in, size_t num);
111
112 #include "md32_common.h"
113
114 #ifdef SHA256_ASM
115 void sha256_block (SHA256_CTX *ctx, const void *in, size_t num, int host);
116 #else
117 static const SHA_LONG K256[64] = {
118         0x428a2f98UL,0x71374491UL,0xb5c0fbcfUL,0xe9b5dba5UL,
119         0x3956c25bUL,0x59f111f1UL,0x923f82a4UL,0xab1c5ed5UL,
120         0xd807aa98UL,0x12835b01UL,0x243185beUL,0x550c7dc3UL,
121         0x72be5d74UL,0x80deb1feUL,0x9bdc06a7UL,0xc19bf174UL,
122         0xe49b69c1UL,0xefbe4786UL,0x0fc19dc6UL,0x240ca1ccUL,
123         0x2de92c6fUL,0x4a7484aaUL,0x5cb0a9dcUL,0x76f988daUL,
124         0x983e5152UL,0xa831c66dUL,0xb00327c8UL,0xbf597fc7UL,
125         0xc6e00bf3UL,0xd5a79147UL,0x06ca6351UL,0x14292967UL,
126         0x27b70a85UL,0x2e1b2138UL,0x4d2c6dfcUL,0x53380d13UL,
127         0x650a7354UL,0x766a0abbUL,0x81c2c92eUL,0x92722c85UL,
128         0xa2bfe8a1UL,0xa81a664bUL,0xc24b8b70UL,0xc76c51a3UL,
129         0xd192e819UL,0xd6990624UL,0xf40e3585UL,0x106aa070UL,
130         0x19a4c116UL,0x1e376c08UL,0x2748774cUL,0x34b0bcb5UL,
131         0x391c0cb3UL,0x4ed8aa4aUL,0x5b9cca4fUL,0x682e6ff3UL,
132         0x748f82eeUL,0x78a5636fUL,0x84c87814UL,0x8cc70208UL,
133         0x90befffaUL,0xa4506cebUL,0xbef9a3f7UL,0xc67178f2UL };
134
135 /*
136  * FIPS specification refers to right rotations, while our ROTATE macro
137  * is left one. This is why you might notice that rotation coefficients
138  * differ from those observed in FIPS document by 32-N...
139  */
140 #define Sigma0(x)       (ROTATE((x),30) ^ ROTATE((x),19) ^ ROTATE((x),10))
141 #define Sigma1(x)       (ROTATE((x),26) ^ ROTATE((x),21) ^ ROTATE((x),7))
142 #define sigma0(x)       (ROTATE((x),25) ^ ROTATE((x),14) ^ ((x)>>3))
143 #define sigma1(x)       (ROTATE((x),15) ^ ROTATE((x),13) ^ ((x)>>10))
144
145 #define Ch(x,y,z)       (((x) & (y)) ^ ((~(x)) & (z)))
146 #define Maj(x,y,z)      (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z)))
147
148 #ifdef OPENSSL_SMALL_FOOTPRINT
149
150 static void sha256_block (SHA256_CTX *ctx, const void *in, size_t num, int host)
151         {
152         unsigned MD32_REG_T a,b,c,d,e,f,g,h,s0,s1,T1,T2;
153         SHA_LONG        X[16],l;
154         int i;
155         const unsigned char *data=in;
156
157                         while (num--) {
158
159         a = ctx->h[0];  b = ctx->h[1];  c = ctx->h[2];  d = ctx->h[3];
160         e = ctx->h[4];  f = ctx->h[5];  g = ctx->h[6];  h = ctx->h[7];
161
162         for (i=0;i<16;i++)
163                 {
164                 HOST_c2l(data,l); T1 = X[i] = l;
165                 T1 += h + Sigma1(e) + Ch(e,f,g) + K256[i];
166                 T2 = Sigma0(a) + Maj(a,b,c);
167                 h = g;  g = f;  f = e;  e = d + T1;
168                 d = c;  c = b;  b = a;  a = T1 + T2;
169                 }
170
171         for (;i<64;i++)
172                 {
173                 s0 = X[(i+1)&0x0f];     s0 = sigma0(s0);
174                 s1 = X[(i+14)&0x0f];    s1 = sigma1(s1);
175
176                 T1 = X[i&0xf] += s0 + s1 + X[(i+9)&0xf];
177                 T1 += h + Sigma1(e) + Ch(e,f,g) + K256[i];
178                 T2 = Sigma0(a) + Maj(a,b,c);
179                 h = g;  g = f;  f = e;  e = d + T1;
180                 d = c;  c = b;  b = a;  a = T1 + T2;
181                 }
182
183         ctx->h[0] += a; ctx->h[1] += b; ctx->h[2] += c; ctx->h[3] += d;
184         ctx->h[4] += e; ctx->h[5] += f; ctx->h[6] += g; ctx->h[7] += h;
185
186                         }
187 }
188
189 #else
190
191 #define ROUND_00_15(i,a,b,c,d,e,f,g,h)          do {    \
192         T1 += h + Sigma1(e) + Ch(e,f,g) + K256[i];      \
193         h = Sigma0(a) + Maj(a,b,c);                     \
194         d += T1;        h += T1;                } while (0)
195
196 #define ROUND_16_63(i,a,b,c,d,e,f,g,h,X)        do {    \
197         s0 = X[(i+1)&0x0f];     s0 = sigma0(s0);        \
198         s1 = X[(i+14)&0x0f];    s1 = sigma1(s1);        \
199         T1 = X[(i)&0x0f] += s0 + s1 + X[(i+9)&0x0f];    \
200         ROUND_00_15(i,a,b,c,d,e,f,g,h);         } while (0)
201
202 static void sha256_block (SHA256_CTX *ctx, const void *in, size_t num, int host)
203         {
204         unsigned MD32_REG_T a,b,c,d,e,f,g,h,s0,s1,T1;
205         SHA_LONG        X[16];
206         int i;
207         const unsigned char *data=in;
208         const union { long one; char little; } is_endian = {1};
209
210                         while (num--) {
211
212         a = ctx->h[0];  b = ctx->h[1];  c = ctx->h[2];  d = ctx->h[3];
213         e = ctx->h[4];  f = ctx->h[5];  g = ctx->h[6];  h = ctx->h[7];
214
215         if (!is_endian.little && sizeof(SHA_LONG)==4 && ((size_t)in%4)==0)
216                 {
217                 const SHA_LONG *W=(const SHA_LONG *)data;
218
219                 T1 = X[0] = W[0];       ROUND_00_15(0,a,b,c,d,e,f,g,h);
220                 T1 = X[1] = W[1];       ROUND_00_15(1,h,a,b,c,d,e,f,g);
221                 T1 = X[2] = W[2];       ROUND_00_15(2,g,h,a,b,c,d,e,f);
222                 T1 = X[3] = W[3];       ROUND_00_15(3,f,g,h,a,b,c,d,e);
223                 T1 = X[4] = W[4];       ROUND_00_15(4,e,f,g,h,a,b,c,d);
224                 T1 = X[5] = W[5];       ROUND_00_15(5,d,e,f,g,h,a,b,c);
225                 T1 = X[6] = W[6];       ROUND_00_15(6,c,d,e,f,g,h,a,b);
226                 T1 = X[7] = W[7];       ROUND_00_15(7,b,c,d,e,f,g,h,a);
227                 T1 = X[8] = W[8];       ROUND_00_15(8,a,b,c,d,e,f,g,h);
228                 T1 = X[9] = W[9];       ROUND_00_15(9,h,a,b,c,d,e,f,g);
229                 T1 = X[10] = W[10];     ROUND_00_15(10,g,h,a,b,c,d,e,f);
230                 T1 = X[11] = W[11];     ROUND_00_15(11,f,g,h,a,b,c,d,e);
231                 T1 = X[12] = W[12];     ROUND_00_15(12,e,f,g,h,a,b,c,d);
232                 T1 = X[13] = W[13];     ROUND_00_15(13,d,e,f,g,h,a,b,c);
233                 T1 = X[14] = W[14];     ROUND_00_15(14,c,d,e,f,g,h,a,b);
234                 T1 = X[15] = W[15];     ROUND_00_15(15,b,c,d,e,f,g,h,a);
235
236                 data += SHA256_CBLOCK;
237                 }
238         else
239                 {
240                 SHA_LONG l;
241
242                 HOST_c2l(data,l); T1 = X[0] = l;  ROUND_00_15(0,a,b,c,d,e,f,g,h);
243                 HOST_c2l(data,l); T1 = X[1] = l;  ROUND_00_15(1,h,a,b,c,d,e,f,g);
244                 HOST_c2l(data,l); T1 = X[2] = l;  ROUND_00_15(2,g,h,a,b,c,d,e,f);
245                 HOST_c2l(data,l); T1 = X[3] = l;  ROUND_00_15(3,f,g,h,a,b,c,d,e);
246                 HOST_c2l(data,l); T1 = X[4] = l;  ROUND_00_15(4,e,f,g,h,a,b,c,d);
247                 HOST_c2l(data,l); T1 = X[5] = l;  ROUND_00_15(5,d,e,f,g,h,a,b,c);
248                 HOST_c2l(data,l); T1 = X[6] = l;  ROUND_00_15(6,c,d,e,f,g,h,a,b);
249                 HOST_c2l(data,l); T1 = X[7] = l;  ROUND_00_15(7,b,c,d,e,f,g,h,a);
250                 HOST_c2l(data,l); T1 = X[8] = l;  ROUND_00_15(8,a,b,c,d,e,f,g,h);
251                 HOST_c2l(data,l); T1 = X[9] = l;  ROUND_00_15(9,h,a,b,c,d,e,f,g);
252                 HOST_c2l(data,l); T1 = X[10] = l; ROUND_00_15(10,g,h,a,b,c,d,e,f);
253                 HOST_c2l(data,l); T1 = X[11] = l; ROUND_00_15(11,f,g,h,a,b,c,d,e);
254                 HOST_c2l(data,l); T1 = X[12] = l; ROUND_00_15(12,e,f,g,h,a,b,c,d);
255                 HOST_c2l(data,l); T1 = X[13] = l; ROUND_00_15(13,d,e,f,g,h,a,b,c);
256                 HOST_c2l(data,l); T1 = X[14] = l; ROUND_00_15(14,c,d,e,f,g,h,a,b);
257                 HOST_c2l(data,l); T1 = X[15] = l; ROUND_00_15(15,b,c,d,e,f,g,h,a);
258                 }
259
260         for (i=16;i<64;i+=8)
261                 {
262                 ROUND_16_63(i+0,a,b,c,d,e,f,g,h,X);
263                 ROUND_16_63(i+1,h,a,b,c,d,e,f,g,X);
264                 ROUND_16_63(i+2,g,h,a,b,c,d,e,f,X);
265                 ROUND_16_63(i+3,f,g,h,a,b,c,d,e,X);
266                 ROUND_16_63(i+4,e,f,g,h,a,b,c,d,X);
267                 ROUND_16_63(i+5,d,e,f,g,h,a,b,c,X);
268                 ROUND_16_63(i+6,c,d,e,f,g,h,a,b,X);
269                 ROUND_16_63(i+7,b,c,d,e,f,g,h,a,X);
270                 }
271
272         ctx->h[0] += a; ctx->h[1] += b; ctx->h[2] += c; ctx->h[3] += d;
273         ctx->h[4] += e; ctx->h[5] += f; ctx->h[6] += g; ctx->h[7] += h;
274
275                         }
276         }
277
278 #endif
279 #endif /* SHA256_ASM */
280
281 void HASH_BLOCK_DATA_ORDER (SHA256_CTX *ctx, const void *in, size_t num)
282 {   sha256_block (ctx,in,num,0);   }
283
284 #endif /* OPENSSL_NO_SHA256 */